A Personalized Recommendation Algorithm based on LSTM Classification

Sentiment analysis is a branch of natural language processing that aims to identify and extract an author’s emotions or emotional attitudes from a text. In today’s society, sentiment analysis is challenged by data bias and misunderstanding, especially in personalized recommendation systems, which may lead to inaccurate information matching with high errors. This paper designs a CBF personalized recommendation model for sentiment analysis by LSTM algorithm, which greatly improves the matching accuracy and recommendation performance. The algorithm reduces the error rate of personalized recommendation of sentiment analysis, solves the problem of information matching error, and the improved accuracy rate is as high as 95.60%. In the future, LSTM algorithms and sentiment analysis personalized recommendation models should be widely used.